Princeton Election Consortium

A first draft of electoral history. Since 2004

New Hampshire: Kasich and Rubio lead a scramble for distant second

February 9th, 2016, 1:00am by Sam Wang


image from http://www.redhotconservative.com/the-musical-chair-of-texas-superintendents/In five polls conducted February 4-8 that include data from the 7th and 8th, after Saturday’s debate, undecideds show up as 8 percent of GOP voters. About 50% of the respondents were reached after the debate. These people aren’t able to express their candidate preference to a pollster, but they do have pretty much all the evidence they are going to get from the candidates themselves. They probably have some preference, but we just can’t find out what it is by asking.

Donald Trump appears to be far in front, at 31%. His lead is 17 percentage points, just about what it’s been since the beginning of August. After that, the undecideds could easily scramble the order of the runners-up… Five candidates have poll medians that are within 8 percentage points of one another: Marco Rubio (14%), John Kasich (13%), Ted Cruz (11%), Jeb Bush (10%), and Chris Christie (6%). (One-sigma uncertainties are plus or minus 0.8 to 2.0 percentage points except for Christie, whose nominal uncertainty is 0.4 percentage point.) These candidates are likely to finish in second though sixth place – but probably not in the order I have listed them.

Marco Rubio faces media blowback from his debate performance, in which Governor Chris Christie trapped Rubio in a loop repeating “let’s dispel with the notion that Obama doesn’t know what he is doing” four times. Even so, Rubio’s support has held steady. He got about a three percentage point bounce the day after the Iowa caucuses. This jump held after Saturday’s debate – totally flat, actually. However, I wonder whether he will pick up many undecideds.

At Stanford’s election law symposium last weekend, Republican macher Ben Ginsberg expressed the general impression that candidates tend to stay in for one round of primaries past the time when it is clear that they should drop out. That would suggest Carly Fiorina and Ben Carson getting out just about now.

However, I am not sure about any of the above-listed non-Trump candidates exiting the race after today. Cruz is in to stay. Kasich has risen slowly since the new year, and a strong finish today would keep him in the race. In fact, I give Kasich an even-odds chance of ending up in second place. Already that’s four candidates – a threshold I have identified as problematic for the Republican Party.

Alternatively, Jeb Bush could pick up a few points, and decide to stick around along with his super-PAC money. Christie’s support is pretty low, but he had so much fun in Saturday’s debate that he will need a pretty hard push to get out. All in all, it seems likely that after New Hampshire, the field will remain divided for a while longer. And since loopholes in the GOP primary rules make delegate selection far from proportional, everyone but Trump will basically continue to be hosed.

On the Democratic side, Bernie Sanders is at a median of 55%, and Hillary Clinton is at 43%. For both of them, the one-sigma uncertainty is 2 percentage points. An upset victory by Clinton looks highly unlikely…but if she held Sanders to a single-digit margin of victory, that would be considered a strong showing in a state that is almost as white as Sanders’s neighboring state of Vermont.

Tags: 2016 Election

27 Comments so far ↓

  • Olav Grinde

    Does anyone know where we can get the fastest and most accurate results from the New Hampshire Presidential Primary?

    In Iowa, the fastest and most accurate results seemed to be the running tallies on the official Democratic and Republican parties’ state websites.

  • Mark F.

    I think if Clinton keeps her losing margin under 10% she will be happy. If Sanders loses (not likely), it’s really over for him.

  • Anthony

    The only way Trump doesn’t get first is if there is some huge systematic error in NH polling. 2nd is completely up for grabs it seems so it kinda validates the non-trump candidates for staying in the race this long, except of course Carson and Fiorina. Although Carson might place 5th or lower I think the only person who quits after NH is Fiorina.

    • Mark F.

      Carson may think he can do better in the South. But I expect Fiorina to drop out as well.

    • Josh

      There’s also the possibility that either A) Trump’s support, in particular, hasn’t been polled correctly, and/or B) that all the late undecideds will break for one candidate.

      I agree that Fiorina will probably bow out either after this or South Carolina; if Chris Christie underperforms, I wouldn’t be surprised if he also leaves the race.

  • bks

    If neither Rubio nor Cruz is in the top 2, the next round is on me. If neither is in the top 3, I’m buying for the entire bar.

    • Josh

      I’ll take that first bet. A shot of Jameson will be fine.

    • JayBoy2k

      I think you are are safe with your 2nd prediction. I would go with that… especially since the votes for the moderates are coming from the same pool.
      On the 1st, you seem to discount the Cruz does not have the ground game in NH and Rubio wounded by repetition arguments..
      You may be right.
      The latest polls imply Bush and Kasich surging toward the finish line.

    • Mark Buckley

      And I’ll have a Bud Light brewed in Merrimack, NH, the bellwether town.

    • whatever next

      Which bar can we find you at?

  • Olav Grinde

    Indeed, as Andrew Prokop eloquently points out in his fascinating article:

    http://www.vox.com/2016/2/8/10938582/marco-rubio-debate

    (NB That’s Vox, not Fox.)

  • JayBoy2k

    I am interested in what the polls and vote distribution can tell us that we do not already know. In Iowa, Trump underperformed his Polls. Is that going to repeat in NH?
    Expectations are that Rubio, Cruz will underperform and Kasich, Bush, Christie will overperform because they have advantages in NH..So the candidates at risk would seem to be these last 3. It is hard to see how any of them would do better in Nevada or SC than in NH.

  • Olav Grinde

    A small question, if I may: Is the Republican Party implementing voter-ID in their New Hampshire primary?
    You know, to make 100 % sure that Bostonians and other infiltrators, as well as Canadians and other aliens, don’t cast illegal votes…

    • Mark Buckley

      Yes Voter ID ( government issued only ) for the first time in NH this year . If you don’t have it you can fill out an affidavit and have your picture attached and then vote

    • Olav Grinde

      Thanks for the update!

  • Amitabh Lath

    I read a statement like “uncertainty is 0.4 percentage point” and immediately red flags go up. These five polls have sample sizes of a few hundred, and all added together N<2000. How do we get sub-1% uncertainties?

    But aside from that I don't think polls can have uncertainties <3%-ish no matter the size, or number of polls aggregated. You cannot measure something down to microns with tape measures, no matter how many you use.

    • dk

      0.4 is 7% of 6. Isn’t that an uncertainty of 7% of Christie’s total.

    • Sam Wang

      That is nominal uncertainty. Not that much smaller than binomial sampling uncertainty, which for 400 respondents per survey, 5 surveys, and a popularity of 6%, would be 0.5%. Recall that uncertainties are smaller for these micro-candidates.

      Down at this level, one could fuss with the confidence intervals more…but really? Not to seem slovenly, but it seemed like overkill considering the uncertainties from polling error and undecided voters.

    • Amitabh Lath

      Thanks, I think I understand. If I wanted to calculate the uncertainty on the difference between say, Bush and Christie, would I add the individual uncertainties in quadrature?

    • Sam Wang

      I would calculate the difference for each survey individually, then do statistics on that.

  • Mark Buckley

    No undecided Evangelicals in New England their preference for Cruz and Rubio already expressed in the polls and they will lose the expectations spin. Look for Boston suburbs to break heavily for Kasich and Bush where moderation is valued more than religious extremism in this very secular state.

  • Olav Grinde

    We need election reform.

    Imagine: all ballots with a “none of the above” option. If that gets the most votes, then new elections are mandated – with the rejected candidates barred from partaking.

    • Todd S. Horowitz

      That’s your election reform?
      How about mandatory voting with instant runoff procedures. Ditch the primaries altogether. Just one big popular vote election.

  • Nicholas Warino

    “Rubio is impossible to predict – but he’s probably lucky the vote wasn’t the day after the debate, which could have seen him down in single digits, whereas he’s probably consolidated since then – or at least put a brake on the downward spiral.”

    How could you know this? Isn’t it just as likely, if not more so, that more time in between the Rubiobot debate and the election means more time for word to spread to the voters, and therefore have a bigger influence on the outcome?

    • whatever next

      I don’t ‘know’ this, none of us do, hence my opinion ‘probably’.

      Your scenario is also plausible, however live viewing of the debate by NH voters is likely to have been huge and reached pretty well all ‘undecideds’ by itself, therefore I came to the conclusion that Rubio’s fleet-footed performances back on the trail would likely mitigate the slip a little / stem the bleed. On balance I think he’s most likely to be 5th now and suggest not higher than 4th.

      It’s all guessing though! It’s worth the risk of egg on my face to have the fun of making a punt on a public forum like this. :)

  • whatever next

    Emerson final poll claims significant drop for Rubio in post-debate day of polling, down to single figures – though his average, and therefore headline numbers, remain in double digits @ 12%.

    Whilst I too still favour Kasich for 2nd place:

    Wouldn’t amaze me if Bush sneaks second.

    Final polls all over the place with four contenders for second, but uniformly suggest Christie is in trouble.

    Rubio is impossible to predict – but he’s probably lucky the vote wasn’t the day after the debate, which could have seen him down in single digits, whereas he’s probably consolidated since then – or at least put a brake on the downward spiral.

  • Mark F.

    Looks like a clusterf–k for 2nd place.

Leave a Comment